Our study has 62 initial participants.
The participants are residents of Gramatneusiedl who have been unemployed for at least 9 months.
This figure shows the characteristics of study participants.
Each picture represents one in ten participants.
| Covariate | Mean wave 1 | Mean wave 2 | Difference | T-statistic | P-value |
|---|---|---|---|---|---|
| Male | 0.581 | 0.581 | 0.000 | 0.000 | 1.000 |
| Age | 44.452 | 44.935 | -0.484 | -0.165 | 0.869 |
| Migration Background | 0.323 | 0.355 | -0.032 | -0.264 | 0.793 |
| Education | 0.452 | 0.452 | 0.000 | 0.000 | 1.000 |
| Medical condition | 0.290 | 0.323 | -0.032 | -0.271 | 0.787 |
| Benefit level | 29.839 | 29.839 | 0.000 | 0.000 | 1.000 |
| Days unemployed | 1721.871 | 1600.839 | 121.032 | 0.483 | 0.631 |
The first component of our study is an experimental comparison between participants who started the job guarantee early, and those who only started later.
We divided participants into pairs that were very similar in terms of their characteristics.
Within each pair, it was randomly decided who would start early (wave 1) and who would start late (wave 2).
We registered this assignment procedure before the study began, to tie our hands: AEA registry
The table compares the average characteristics of participants of the two waves.
Because of our procedure, they are very similar.
t-statistics and p-values are for “naive” inference, ignoring the pairwise matching procedure for treatment assignment.
The second component of our study is a comparison of Gramatneusiedl (the treated municipality) with other, similar municipalities in lower Austria.
These municipalities were automatically selected based on their labor market and economic characteristics, such that their weighted average looks like Gramatneusiedl before the program started.
This map shows these control municipalities and their weights.
Brighter dots correspond to municipalities with a larger weight in the synthetic control.
Hover over the dots to see their names and weights.
| Weight | Municipality |
|---|---|
| 0.487 | Ebreichsdorf |
| 0.203 | Zeillern |
| 0.134 | Rußbach |
| 0.079 | Leopoldsdorf im Marchfelde |
| 0.046 | Strasshof an der Nordbahn |
| 0.024 | Sieghartskirchen |
| 0.023 | Sollenau |
This table displays the same information as the map on the previous tab.
It lists the synthetic control municipalities sorted by their weights.
We again registered these weights before the study began, to tie our hands: AEA registry
This figure shows average outcomes in different categories, for three groups of people:
They are shown in the figure as follows:
Higher values imply better outcomes. Outcomes are scaled to have range 0 to 1.
Each outcome is an index based on several survey questions.
The exact survey questions can be found here: AEA registry - survey
This plot shows confidence intervals for the population average treatment effect.
95% confidence intervals are represented by a thin line; 90% confidence intervals by a thick line.
Estimates are calculated based on the subsample of complete pairs.